Path: blob/master/thirdparty/icu4c/common/brkiter.cpp
21026 views
// © 2016 and later: Unicode, Inc. and others.1// License & terms of use: http://www.unicode.org/copyright.html2/*3*******************************************************************************4* Copyright (C) 1997-2015, International Business Machines Corporation and5* others. All Rights Reserved.6*******************************************************************************7*8* File brkiter.cpp9*10* Modification History:11*12* Date Name Description13* 02/18/97 aliu Converted from OpenClass. Added DONE.14* 01/13/2000 helena Added UErrorCode parameter to createXXXInstance methods.15*****************************************************************************************16*/1718// *****************************************************************************19// This file was generated from the java source file BreakIterator.java20// *****************************************************************************2122#include "unicode/utypes.h"2324#if !UCONFIG_NO_BREAK_ITERATION2526#include "unicode/rbbi.h"27#include "unicode/brkiter.h"28#include "unicode/udata.h"29#include "unicode/uloc.h"30#include "unicode/ures.h"31#include "unicode/ustring.h"32#include "unicode/filteredbrk.h"33#include "bytesinkutil.h"34#include "ucln_cmn.h"35#include "cstring.h"36#include "umutex.h"37#include "servloc.h"38#include "locbased.h"39#include "uresimp.h"40#include "uassert.h"41#include "ubrkimpl.h"42#include "utracimp.h"43#include "charstr.h"4445// *****************************************************************************46// class BreakIterator47// This class implements methods for finding the location of boundaries in text.48// Instances of BreakIterator maintain a current position and scan over text49// returning the index of characters where boundaries occur.50// *****************************************************************************5152U_NAMESPACE_BEGIN5354// -------------------------------------5556BreakIterator*57BreakIterator::buildInstance(const Locale& loc, const char *type, UErrorCode &status)58{59char fnbuff[256];60char ext[4]={'\0'};61CharString actual;62int32_t size;63const char16_t* brkfname = nullptr;64UResourceBundle brkRulesStack;65UResourceBundle brkNameStack;66UResourceBundle *brkRules = &brkRulesStack;67UResourceBundle *brkName = &brkNameStack;68RuleBasedBreakIterator *result = nullptr;6970if (U_FAILURE(status))71return nullptr;7273ures_initStackObject(brkRules);74ures_initStackObject(brkName);7576// Get the locale77UResourceBundle *b = ures_openNoDefault(U_ICUDATA_BRKITR, loc.getName(), &status);7879// Get the "boundaries" array.80if (U_SUCCESS(status)) {81brkRules = ures_getByKeyWithFallback(b, "boundaries", brkRules, &status);82// Get the string object naming the rules file83brkName = ures_getByKeyWithFallback(brkRules, type, brkName, &status);84// Get the actual string85brkfname = ures_getString(brkName, &size, &status);86U_ASSERT((size_t)size<sizeof(fnbuff));87if (static_cast<size_t>(size) >= sizeof(fnbuff)) {88size=0;89if (U_SUCCESS(status)) {90status = U_BUFFER_OVERFLOW_ERROR;91}92}9394// Use the string if we found it95if (U_SUCCESS(status) && brkfname) {96actual.append(ures_getLocaleInternal(brkName, &status), -1, status);9798char16_t* extStart=u_strchr(brkfname, 0x002e);99int len = 0;100if (extStart != nullptr){101len = static_cast<int>(extStart - brkfname);102u_UCharsToChars(extStart+1, ext, sizeof(ext)); // nul terminates the buff103u_UCharsToChars(brkfname, fnbuff, len);104}105fnbuff[len]=0; // nul terminate106}107}108109ures_close(brkRules);110ures_close(brkName);111112UDataMemory* file = udata_open(U_ICUDATA_BRKITR, ext, fnbuff, &status);113if (U_FAILURE(status)) {114ures_close(b);115return nullptr;116}117118// Create a RuleBasedBreakIterator119result = new RuleBasedBreakIterator(file, uprv_strstr(type, "phrase") != nullptr, status);120121// If there is a result, set the valid locale and actual locale, and the kind122if (U_SUCCESS(status) && result != nullptr) {123result->actualLocale = Locale(actual.data());124result->validLocale = Locale(ures_getLocaleByType(b, ULOC_VALID_LOCALE, &status));125result->requestLocale = loc;126}127128ures_close(b);129130if (U_FAILURE(status) && result != nullptr) { // Sometimes redundant check, but simple131delete result;132return nullptr;133}134135if (result == nullptr) {136udata_close(file);137if (U_SUCCESS(status)) {138status = U_MEMORY_ALLOCATION_ERROR;139}140}141142return result;143}144145// Creates a break iterator for word breaks.146BreakIterator* U_EXPORT2147BreakIterator::createWordInstance(const Locale& key, UErrorCode& status)148{149return createInstance(key, UBRK_WORD, status);150}151152// -------------------------------------153154// Creates a break iterator for line breaks.155BreakIterator* U_EXPORT2156BreakIterator::createLineInstance(const Locale& key, UErrorCode& status)157{158return createInstance(key, UBRK_LINE, status);159}160161// -------------------------------------162163// Creates a break iterator for character breaks.164BreakIterator* U_EXPORT2165BreakIterator::createCharacterInstance(const Locale& key, UErrorCode& status)166{167return createInstance(key, UBRK_CHARACTER, status);168}169170// -------------------------------------171172// Creates a break iterator for sentence breaks.173BreakIterator* U_EXPORT2174BreakIterator::createSentenceInstance(const Locale& key, UErrorCode& status)175{176return createInstance(key, UBRK_SENTENCE, status);177}178179// -------------------------------------180181// Creates a break iterator for title casing breaks.182BreakIterator* U_EXPORT2183BreakIterator::createTitleInstance(const Locale& key, UErrorCode& status)184{185return createInstance(key, UBRK_TITLE, status);186}187188// -------------------------------------189190// Gets all the available locales that has localized text boundary data.191const Locale* U_EXPORT2192BreakIterator::getAvailableLocales(int32_t& count)193{194return Locale::getAvailableLocales(count);195}196197// ------------------------------------------198//199// Constructors, destructor and assignment operator200//201//-------------------------------------------202203BreakIterator::BreakIterator()204: actualLocale(Locale::getRoot()), validLocale(Locale::getRoot()), requestLocale(Locale::getRoot())205{206}207208BreakIterator::BreakIterator(const BreakIterator &other)209: UObject(other),210actualLocale(other.actualLocale),211validLocale(other.validLocale),212requestLocale(other.requestLocale) {213}214215BreakIterator &BreakIterator::operator =(const BreakIterator &other) {216if (this != &other) {217actualLocale = other.actualLocale;218validLocale = other.validLocale;219requestLocale = other.requestLocale;220}221return *this;222}223224BreakIterator::~BreakIterator()225{226}227228// ------------------------------------------229//230// Registration231//232//-------------------------------------------233#if !UCONFIG_NO_SERVICE234235// -------------------------------------236237class ICUBreakIteratorFactory : public ICUResourceBundleFactory {238public:239virtual ~ICUBreakIteratorFactory();240protected:241virtual UObject* handleCreate(const Locale& loc, int32_t kind, const ICUService* /*service*/, UErrorCode& status) const override {242return BreakIterator::makeInstance(loc, kind, status);243}244};245246ICUBreakIteratorFactory::~ICUBreakIteratorFactory() {}247248// -------------------------------------249250class ICUBreakIteratorService : public ICULocaleService {251public:252ICUBreakIteratorService()253: ICULocaleService(UNICODE_STRING("Break Iterator", 14))254{255UErrorCode status = U_ZERO_ERROR;256registerFactory(new ICUBreakIteratorFactory(), status);257}258259virtual ~ICUBreakIteratorService();260261virtual UObject* cloneInstance(UObject* instance) const override {262return ((BreakIterator*)instance)->clone();263}264265virtual UObject* handleDefault(const ICUServiceKey& key, UnicodeString* /*actualID*/, UErrorCode& status) const override {266LocaleKey& lkey = static_cast<LocaleKey&>(const_cast<ICUServiceKey&>(key));267int32_t kind = lkey.kind();268Locale loc;269lkey.currentLocale(loc);270return BreakIterator::makeInstance(loc, kind, status);271}272273virtual UBool isDefault() const override {274return countFactories() == 1;275}276};277278ICUBreakIteratorService::~ICUBreakIteratorService() {}279280// -------------------------------------281282// defined in ucln_cmn.h283U_NAMESPACE_END284285static icu::UInitOnce gInitOnceBrkiter {};286static icu::ICULocaleService* gService = nullptr;287288289290/**291* Release all static memory held by breakiterator.292*/293U_CDECL_BEGIN294static UBool U_CALLCONV breakiterator_cleanup() {295#if !UCONFIG_NO_SERVICE296if (gService) {297delete gService;298gService = nullptr;299}300gInitOnceBrkiter.reset();301#endif302return true;303}304U_CDECL_END305U_NAMESPACE_BEGIN306307static void U_CALLCONV308initService() {309gService = new ICUBreakIteratorService();310ucln_common_registerCleanup(UCLN_COMMON_BREAKITERATOR, breakiterator_cleanup);311}312313static ICULocaleService*314getService()315{316umtx_initOnce(gInitOnceBrkiter, &initService);317return gService;318}319320321// -------------------------------------322323static inline UBool324hasService()325{326return !gInitOnceBrkiter.isReset() && getService() != nullptr;327}328329// -------------------------------------330331URegistryKey U_EXPORT2332BreakIterator::registerInstance(BreakIterator* toAdopt, const Locale& locale, UBreakIteratorType kind, UErrorCode& status)333{334ICULocaleService *service = getService();335if (service == nullptr) {336status = U_MEMORY_ALLOCATION_ERROR;337return nullptr;338}339return service->registerInstance(toAdopt, locale, kind, status);340}341342// -------------------------------------343344UBool U_EXPORT2345BreakIterator::unregister(URegistryKey key, UErrorCode& status)346{347if (U_SUCCESS(status)) {348if (hasService()) {349return gService->unregister(key, status);350}351status = U_MEMORY_ALLOCATION_ERROR;352}353return false;354}355356// -------------------------------------357358StringEnumeration* U_EXPORT2359BreakIterator::getAvailableLocales()360{361ICULocaleService *service = getService();362if (service == nullptr) {363return nullptr;364}365return service->getAvailableLocales();366}367#endif /* UCONFIG_NO_SERVICE */368369// -------------------------------------370371BreakIterator*372BreakIterator::createInstance(const Locale& loc, int32_t kind, UErrorCode& status)373{374if (U_FAILURE(status)) {375return nullptr;376}377378#if !UCONFIG_NO_SERVICE379if (hasService()) {380Locale actualLoc("");381BreakIterator *result = (BreakIterator*)gService->get(loc, kind, &actualLoc, status);382// TODO: The way the service code works in ICU 2.8 is that if383// there is a real registered break iterator, the actualLoc384// will be populated, but if the handleDefault path is taken385// (because nothing is registered that can handle the386// requested locale) then the actualLoc comes back empty. In387// that case, the returned object already has its actual/valid388// locale data populated (by makeInstance, which is what389// handleDefault calls), so we don't touch it. YES, A COMMENT390// THIS LONG is a sign of bad code -- so the action item is to391// revisit this in ICU 3.0 and clean it up/fix it/remove it.392if (U_SUCCESS(status) && (result != nullptr) && *actualLoc.getName() != 0) {393result->actualLocale = actualLoc;394result->validLocale = actualLoc;395}396return result;397}398else399#endif400{401return makeInstance(loc, kind, status);402}403}404405// -------------------------------------406enum { kKeyValueLenMax = 32 };407408BreakIterator*409BreakIterator::makeInstance(const Locale& loc, int32_t kind, UErrorCode& status)410{411412if (U_FAILURE(status)) {413return nullptr;414}415416BreakIterator *result = nullptr;417switch (kind) {418case UBRK_CHARACTER:419{420UTRACE_ENTRY(UTRACE_UBRK_CREATE_CHARACTER);421result = BreakIterator::buildInstance(loc, "grapheme", status);422UTRACE_EXIT_STATUS(status);423}424break;425case UBRK_WORD:426{427UTRACE_ENTRY(UTRACE_UBRK_CREATE_WORD);428result = BreakIterator::buildInstance(loc, "word", status);429UTRACE_EXIT_STATUS(status);430}431break;432case UBRK_LINE:433{434char lb_lw[kKeyValueLenMax];435UTRACE_ENTRY(UTRACE_UBRK_CREATE_LINE);436uprv_strcpy(lb_lw, "line");437UErrorCode kvStatus = U_ZERO_ERROR;438auto value = loc.getKeywordValue<CharString>("lb", kvStatus);439if (U_SUCCESS(kvStatus) && (value == "strict" || value == "normal" || value == "loose")) {440uprv_strcat(lb_lw, "_");441uprv_strcat(lb_lw, value.data());442}443// lw=phrase is only supported in Japanese and Korean444if (uprv_strcmp(loc.getLanguage(), "ja") == 0 || uprv_strcmp(loc.getLanguage(), "ko") == 0) {445value = loc.getKeywordValue<CharString>("lw", kvStatus);446if (U_SUCCESS(kvStatus) && value == "phrase") {447uprv_strcat(lb_lw, "_");448uprv_strcat(lb_lw, value.data());449}450}451result = BreakIterator::buildInstance(loc, lb_lw, status);452453UTRACE_DATA1(UTRACE_INFO, "lb_lw=%s", lb_lw);454UTRACE_EXIT_STATUS(status);455}456break;457case UBRK_SENTENCE:458{459UTRACE_ENTRY(UTRACE_UBRK_CREATE_SENTENCE);460result = BreakIterator::buildInstance(loc, "sentence", status);461#if !UCONFIG_NO_FILTERED_BREAK_ITERATION462char ssKeyValue[kKeyValueLenMax] = {0};463UErrorCode kvStatus = U_ZERO_ERROR;464int32_t kLen = loc.getKeywordValue("ss", ssKeyValue, kKeyValueLenMax, kvStatus);465if (U_SUCCESS(kvStatus) && kLen > 0 && uprv_strcmp(ssKeyValue,"standard")==0) {466FilteredBreakIteratorBuilder* fbiBuilder = FilteredBreakIteratorBuilder::createInstance(loc, kvStatus);467if (U_SUCCESS(kvStatus)) {468result = fbiBuilder->build(result, status);469delete fbiBuilder;470}471}472#endif473UTRACE_EXIT_STATUS(status);474}475break;476case UBRK_TITLE:477{478UTRACE_ENTRY(UTRACE_UBRK_CREATE_TITLE);479result = BreakIterator::buildInstance(loc, "title", status);480UTRACE_EXIT_STATUS(status);481}482break;483default:484status = U_ILLEGAL_ARGUMENT_ERROR;485}486487if (U_FAILURE(status)) {488delete result;489return nullptr;490}491492return result;493}494495Locale496BreakIterator::getLocale(ULocDataLocaleType type, UErrorCode& status) const {497if (U_FAILURE(status)) {498return Locale::getRoot();499}500if (type == ULOC_REQUESTED_LOCALE) {501return requestLocale;502}503return LocaleBased::getLocale(validLocale, actualLocale, type, status);504}505506const char *507BreakIterator::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const {508if (U_FAILURE(status)) {509return nullptr;510}511if (type == ULOC_REQUESTED_LOCALE) {512return requestLocale.getName();513}514return LocaleBased::getLocaleID(validLocale, actualLocale, type, status);515}516517518// This implementation of getRuleStatus is a do-nothing stub, here to519// provide a default implementation for any derived BreakIterator classes that520// do not implement it themselves.521int32_t BreakIterator::getRuleStatus() const {522return 0;523}524525// This implementation of getRuleStatusVec is a do-nothing stub, here to526// provide a default implementation for any derived BreakIterator classes that527// do not implement it themselves.528int32_t BreakIterator::getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status) {529if (U_FAILURE(status)) {530return 0;531}532if (capacity < 1) {533status = U_BUFFER_OVERFLOW_ERROR;534return 1;535}536*fillInVec = 0;537return 1;538}539540BreakIterator::BreakIterator(const Locale& valid, const Locale& actual)541: actualLocale(actual), validLocale(valid), requestLocale(Locale::getRoot()) {542}543544U_NAMESPACE_END545546#endif /* #if !UCONFIG_NO_BREAK_ITERATION */547548//eof549550551